This replication package provides code, selected data, and pseudodata for the article "Poison Pills in the Shadow of the Law" by Martijn Cremers, Lubomir Litov, Simone Sepe, and Michał Zator.
Prepared: August 14, 2024

===================================

The paper relies on several datasets and some of them are confidential. All data that we are allowed to share is included in the "Source Data" folder (e.g., selected incorporation data, law passage data). The beginning of PoisonPillsInTheShadowOfTheLaw_JFQA.do file contains a list of datasets required to create the main analysis file, "PPL_toanalyze.dta", including those that we are not allowed to share (e.g., Compustat, ISS, etc.), so that interested researchers can obtain these datasets on their own and replicate all the findings using the attached code. 

The main code file is PoisonPillsInTheShadowOfTheLaw_JFQA.do, which loads the data and generates results for almost all exhibits in the paper, excluding Table OA10 and part of Table 5. The original data is merged and preprocessed in lines 32-368 of the main code, which produces "PPL_toanalyze.dta". While we are unable to include the original PPL_toanalyze.dta file, as it contains data that we are not allowed to share, we include a pseudo-data version of PPL_toanalyze.dta. The data structure of this file is the same as of the original file, but numerical values are not meaningful. As a result, the attache code run with pseudo-data is not going to reproduce meaningful results. In addition, the structure of some supplementary datasets created in the analysis process will also not be meaningful (e.g., in the matching analysis, matches won't be identified for some observations). Hence, attached pseudodata should be viewed only as a way to facilitate testing of the replication code.

====================================

To generate equivalents of all exhibits of the paper based on pseudodata:

1) To generate the majority of exhibits, change line 3 in PoisonPillsInTheShadowOfTheLaw_JFQA.do to the path where this file is located and execute the code in Stata. This will generate all results in the paper and in the online appendix, excluding Table OA10 and column 6 of Table 5, Panel A, and columns 4-6 of Table 5, Panel B. Column 6, Table 5, Panel A uses an alternative definition of the poison pill, which effectively changes the construction of many variables used in the paper. Therefore, to generate this single specification one needs to re-run the main code after uncommenting a few lines of the code as described in lines 773-777. Please note that running the code with the pseudodata renders some results 

2) To generate stacked regression results from Table 5B, please run Stacked Regressions.do file after changing the line 3 to the file location.

3) Table OA10 is generated using Eventus (part of WRDS). The content of folder "TableOA10" includes details of the procedure used (In Table OA10 Read Me.txt) and the data & code (Table_OA10.do) needed to generate the final results. Please note that we include the file CARs.dta, which is enough to generate the actual results in the table. All other steps and datasets are needed to generate CARs.dta, with the help of Eventus.